_version_ 1866912767022202880
author Noori, Ayush
Polonuer, Joaquín
Meyer, Katharina
Budnik, Bogdan
Morton, Shad
Wang, Xinyuan
Nazeen, Sumaiya
He, Yingnan
Arango, Iñaki
Vittor, Lucas
Woodworth, Matthew
Krolewski, Richard C.
Li, Michelle M.
Liu, Ninning
Kamath, Tushar
Macosko, Evan
Ritter, Dylan
Afroz, Jalwa
Henderson, Alexander B. H.
Studer, Lorenz
Rodriques, Samuel G.
White, Andrew
Dagan, Noa
Clifton, David A.
Church, George M.
Das, Sudeshna
Tam, Jenny M.
Khurana, Vikram
Zitnik, Marinka
author_facet Noori, Ayush
Polonuer, Joaquín
Meyer, Katharina
Budnik, Bogdan
Morton, Shad
Wang, Xinyuan
Nazeen, Sumaiya
He, Yingnan
Arango, Iñaki
Vittor, Lucas
Woodworth, Matthew
Krolewski, Richard C.
Li, Michelle M.
Liu, Ninning
Kamath, Tushar
Macosko, Evan
Ritter, Dylan
Afroz, Jalwa
Henderson, Alexander B. H.
Studer, Lorenz
Rodriques, Samuel G.
White, Andrew
Dagan, Noa
Clifton, David A.
Church, George M.
Das, Sudeshna
Tam, Jenny M.
Khurana, Vikram
Zitnik, Marinka
contents Neurological diseases are the leading global cause of disability, yet most lack disease-modifying treatments. We present PROTON, a heterogeneous graph transformer that generates testable hypotheses across molecular, organoid, and clinical systems. To evaluate PROTON, we apply it to Parkinson's disease (PD), bipolar disorder (BD), and Alzheimer's disease (AD). In PD, PROTON linked genetic risk loci to genes essential for dopaminergic neuron survival and predicted pesticides toxic to patient-derived neurons, including the insecticide endosulfan, which ranked within the top 1.29% of predictions. In silico screens performed by PROTON reproduced six genome-wide $α$-synuclein experiments, including a split-ubiquitin yeast two-hybrid system (normalized enrichment score [NES] = 2.30, FDR-adjusted $p < 1 \times 10^{-4}$), an ascorbate peroxidase proximity labeling assay (NES = 2.16, FDR $< 1 \times 10^{-4}$), and a high-depth targeted exome sequencing study in 496 synucleinopathy patients (NES = 2.13, FDR $< 1 \times 10^{-4}$). In BD, PROTON predicted calcitriol as a candidate drug that reversed proteomic alterations observed in cortical organoids derived from BD patients. In AD, we evaluated PROTON predictions in health records from $n = 610,524$ patients at Mass General Brigham, confirming that five PROTON-predicted drugs were associated with reduced seven-year dementia risk (minimum hazard ratio = 0.63, 95% CI: 0.53-0.75, $p < 1 \times 10^{-7}$). PROTON generated neurological hypotheses that were evaluated across molecular, organoid, and clinical systems, defining a path for AI-driven discovery in neurological disease.
format Preprint
id arxiv_https___arxiv_org_abs_2512_13724
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Graph AI generates neurological hypotheses validated in molecular, organoid, and clinical systems
Noori, Ayush
Polonuer, Joaquín
Meyer, Katharina
Budnik, Bogdan
Morton, Shad
Wang, Xinyuan
Nazeen, Sumaiya
He, Yingnan
Arango, Iñaki
Vittor, Lucas
Woodworth, Matthew
Krolewski, Richard C.
Li, Michelle M.
Liu, Ninning
Kamath, Tushar
Macosko, Evan
Ritter, Dylan
Afroz, Jalwa
Henderson, Alexander B. H.
Studer, Lorenz
Rodriques, Samuel G.
White, Andrew
Dagan, Noa
Clifton, David A.
Church, George M.
Das, Sudeshna
Tam, Jenny M.
Khurana, Vikram
Zitnik, Marinka
Quantitative Methods
Artificial Intelligence
Neurons and Cognition
Neurological diseases are the leading global cause of disability, yet most lack disease-modifying treatments. We present PROTON, a heterogeneous graph transformer that generates testable hypotheses across molecular, organoid, and clinical systems. To evaluate PROTON, we apply it to Parkinson's disease (PD), bipolar disorder (BD), and Alzheimer's disease (AD). In PD, PROTON linked genetic risk loci to genes essential for dopaminergic neuron survival and predicted pesticides toxic to patient-derived neurons, including the insecticide endosulfan, which ranked within the top 1.29% of predictions. In silico screens performed by PROTON reproduced six genome-wide $α$-synuclein experiments, including a split-ubiquitin yeast two-hybrid system (normalized enrichment score [NES] = 2.30, FDR-adjusted $p < 1 \times 10^{-4}$), an ascorbate peroxidase proximity labeling assay (NES = 2.16, FDR $< 1 \times 10^{-4}$), and a high-depth targeted exome sequencing study in 496 synucleinopathy patients (NES = 2.13, FDR $< 1 \times 10^{-4}$). In BD, PROTON predicted calcitriol as a candidate drug that reversed proteomic alterations observed in cortical organoids derived from BD patients. In AD, we evaluated PROTON predictions in health records from $n = 610,524$ patients at Mass General Brigham, confirming that five PROTON-predicted drugs were associated with reduced seven-year dementia risk (minimum hazard ratio = 0.63, 95% CI: 0.53-0.75, $p < 1 \times 10^{-7}$). PROTON generated neurological hypotheses that were evaluated across molecular, organoid, and clinical systems, defining a path for AI-driven discovery in neurological disease.
title Graph AI generates neurological hypotheses validated in molecular, organoid, and clinical systems
topic Quantitative Methods
Artificial Intelligence
Neurons and Cognition
url https://arxiv.org/abs/2512.13724